This project focuses on the innovative application of statistical methods to analyze data on human illnesses, particularly cancer. Most of our efforts centered on data from a large national study. We analyzed data from the Surveillance, Epidemiology, and End Results (SEER) study to assess long-term trends in cancer incidence and deviations from those trends attributable to birth cohorts or to calendar periods. We analyzed data collected from 1975 to 1994 as part of the SEER study, which represents about 10% of the US population. Our goal was to search for and to draw attention to unexplained patterns in cancer incidence. The analysis adjusted for age, time period, and birth cohort. We focused on long-term trends and on deviations from those trends attributable to birth cohorts or to calendar periods. On average, cancer incidence rose 0.8% annually in white women and 1.8% in white men. After removing several cancers related to smoking and increased screening, average annual increases fell to 0.1% in white women but persisted at 1.7% in white men. In particular, yearly increases in non-Hodgkins lymphoma averaged 2.4% in white women and 4.7% in white men. Among men, incidence changes attributable to cohorts grew progressively larger from one cohort to the next. Cancer incidence patterns among black men and women were similar to those among whites despite smaller population sizes. - SEER, smoking, screening, trends, cancer incidence, age-period-cohort analysis